Stochastic approach in solving kinestatic equations for constrained robot control

摘要

The kinestatic filtering problem consists in inverting the equations that describe the constrained robot kinematics and statics (kinestatic equations), and its objective is to obtain from noisy measures clean position and force information to be used in a hybrid control scheme. Almost all researchers in the field have focused their attention on the deterministic formulation of the problem, and in particular on its solution by pseudoinversion. This work shortly reviews deterministic approach and investigates the use of stochastic tools to develop kinestatic filters, so to compare features and properties of both approaches and to discuss the physical meaning of the so-lutions. An interesting point is that the stochastic approach allows to exploit the correlation among measurements at consecutive sampling instants in the construction of recursive kinestatic filters.